REDUCING PROGRAM DISTURBS IN NON-VOLATILE MEMORY CELLS

Description

[0001] The present disclosure relates generally to memory devices, and more particularly to methods for reducing program disturbs in non-volatile memory cells.

[0002] US 2006/083066 A1 discloses a semiconductor device comprising a memory cell array and a source line driver. Each of the memory cells in the memory cell array has a floating gate cell transistor which stores data by accumulating charge in the floating gate and a select gate transistor whose drain is connected to the source of the cell transistor and whose source is connected to a source line. The source line driver is configured so as to drive the source line in a write operation at a potential between the substrate bias potential of the cell transistor and select gate transistor and the ground potential.

[0003] US2003/177301 A1 discloses a method for operating a non-volatile memory comprising an array of cells having a memory transistor and a select transistor.

[0004] Non-volatile memories are widely used for storing data in computer systems, and typically include a memory array with a large number of memory cells arranged in rows and columns. Each of the memory cells includes a non-volatile charge trapping gate field-effect transistor that is programmed or erased by applying a voltage of the proper polarity, magnitude and duration between a control gate and the substrate. A positive gate-to-substrate voltage causes electrons to tunnel from the channel to a charge-trapping dielectric layer raising a threshold voltage (V_T) of the transistor, and a negative gate-to-channel voltage causes holes to tunnel from the channel to the charge-trapping dielectric layer lowering the threshold voltage.

[0005] Non-volatile memories suffer from program or bitline disturbs, which is an unintended and detrimental change in memory cell V_T when another memory cell connected to the same bitline is inhibited from being programmed. Bitline disturb refers to disturb of the memory cells located in a row different from the row containing the cell undergoing programming. Bitline disturb occurring in the deselected row increases as the number of erase/program cycles in rows selected in the common well increases. The magnitude of bitline disturb also increases at higher temperatures, and, since memory cell dimensions scale down faster than applied voltages at advanced technology nodes, bitline disturb also becomes worse as the density of non-volatile memories increase.

[0006] It is, therefore, an object of the present invention to provide improved non-volatile memories and methods of programming the same.

[0007] This object is achieved with the features of the claims.

BRIEF DESCRIPTION OF THE DRAWINGS

[0008] The present invention will be understood more fully from the detailed description that follows and from the accompanying drawings and the appended claims provided below, where:

FIG. 1 is a block diagram illustrating a cross-sectional side view of a non-volatile memory transistor or device;

FIG. 2 is a schematic diagram illustrating a two transistor (2T) memory cell for which an embodiment of the present disclosure is particularly useful;

FIG. 3 is a schematic diagram is a segment of a memory array illustrating an embodiment of a program operation according to the present disclosure;

FIG. 4 is a graph illustrating a positive high voltage (V_POS), a negative high voltage (V_NEG), and an intermediate, margin voltage (V_MARG) according to an embodiment of the present disclosure;

FIG. 5 is a graph illustrating voltages applied to a selected global wordline (V_{SELECTED WL}) and a deselected global wordline (V_{DESELECTED GWL}) during a program operation according to an embodiment of the present disclosure;

FIG. 6 is a block diagram illustrating a processing system including a memory device according to an embodiment of the present disclosure;

FIGs. 7A-7C are block diagrams illustrating details of command and control circuitry of a non-volatile memory according to various embodiments of the present disclosure; and

FIG. 8 is a flowchart illustrating a method for reducing bitline disturbs in unselected memory cells according to an embodiment of the present disclosure.

DETAILED DESCRIPTION

[0009] Methods for reducing program disturbs in non-volatile memories are described herein. The method is particularly useful for operating memories made of memory arrays of bit cells or memory cells including non-volatile trapped-charge semiconductor devices that may be programmed or erased by applying a voltage of the proper polarity, magnitude and duration.

[0010] In the following description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the present invention. It will be evident, however, to one skilled in the art that the present invention may be practiced without these specific details. In other instances, well-known structures, and techniques are not shown in detail or are shown in block diagram form in order to avoid unnecessarily obscuring an understanding of this description.

[0011] Reference in the description to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the invention. The appearances of the phrase "in one embodiment" in various places in the specification do not necessarily all refer to the same embodiment. The term to couple as used herein may include both to directly electrically connect two or more components or elements and to indirectly connect through one or more intervening components

[0012] The non-volatile memory may include memory cells with a non-volatile memory transistor or device implemented using Silicon-Oxide-Nitride-Oxide-Silicon (SONOS) or floating gate technology.

[0013] In one embodiment, illustrated in FIG. 1, the non-volatile memory transistor or device is a SONOS-type non-volatile memory device. Referring to FIG. 1, a SONOS device 100 includes a gate stack 102 formed over a substrate 104. The SONOS device 100 further includes source/drain regions 106 formed in a well 108 in the substrate 104 on either side of gate stack 102, which define a channel region 110 underneath gate stack. Gate stack 102 includes an oxide tunnel dielectric layer 112, a nitride or oxynitride charge-trapping layer 114, a top, blocking oxide layer 116 and a poly-silicon (poly) or metal layer which serves as a control gate 118.

[0014] When the control gate 118 is appropriately biased, electrons from the source/drain regions 106 are injected or tunnel through tunnel dielectric layer 112 and are trapped in the charge-trapping layer 114. The mechanisms by which charge is injected can include both Fowler-Nordheim (FN) tunneling and hot-carrier injection. The charge trapped in the charge-trapping layer 114 results in an energy barrier between the drain and the source, raising the threshold voltage V_T necessary to turn on the SONOS device 100 putting the device in a "programmed" state. The SONOS device 100 can be "erased" or the trapped charge removed and replaced with holes by applying an opposite bias on the control gate 118.

[0015] In another embodiment, the non-volatile trapped-charge semiconductor device can be a floating-gate MOS field-effect transistor (FGMOS) or device. Generally, is similar in structure to the SONOS device 100 described above, differing primarily in that a FGMOS includes a poly-silicon (poly) floating gate, which is capacitively coupled to inputs of the device, rather than a nitride or oxynitride charge-trapping. Thus, the FGMOS device can be described with reference to FIG. 1. Referring to FIG. 1, a FGMOS device 100 includes a gate stack 102 formed over a substrate 104. The FGMOS device 100 further includes source/drain regions 106 formed in a well 108 in the substrate 104 on either side of gate stack 102, which define a channel region 110 underneath gate stack. Gate stack 102 includes a tunnel dielectric layer 112, a floating gate layer 114, a blocking oxide or top dielectric layer 116 and a poly-silicon or metal layer which serves as a control gate 118.

[0016] Similarly to the SONOS device described above the FGMOS device 100 can be programmed by applying an appropriate bias between the control gate and the source and drain regions to inject charge in to the charge-trapping layer, raising the threshold voltage V_T necessary to turn on the FGMOS device. The FGMOS device can be erased or the trapped charge removed by applying an opposite bias on the control gate.

[0017] A memory array is constructed by fabricating a grid of memory cells arranged in rows and columns and connected by a number of horizontal and vertical control lines to peripheral circuitry such as address decoders and sense amplifiers. Each memory cell includes at least one non-volatile trapped-charge semiconductor device, such as those described above, and may have a one transistor (1T) or two transistor (2T) architecture.

[0018] In one embodiment, illustrated in FIG. 2, the memory cell 200 has a 2T-architecture and includes, in addition to a non-volatile memory transistor 202, a pass or select transistor 204, for example, a conventional IGFET sharing a common substrate connection 206 with the memory transistor 202. Referring to FIG. 2, the memory transistor 202 has a charge trapping layer 208 and a drain 210 connected to a source 222 of the select transistor 204 and through the select transistor to a bitline 212, a control gate 214 connected to a wordline 216 and a source 218 connected to a source line 224. Select transistor 204 also includes a drain 220 connected to a bitline 212 and a gate 226 connected to a select or read line 228.

[0019] During an erase operation to erase the memory cell 200 a negative high voltage (V_NEG) is applied to the wordline 216 and a positive high voltage (V_POS) applied to the bitline and the substrate connection 206. Generally, the memory cell 200 is erased as part of a bulk erase operation in which all memory cells in a selected row of a memory array are erased at once prior to a program operation to program the memory cell 200 by applying the appropriate voltages to a global wordline (GWL) shared by all memory cells in the row, the substrate connection and to all bitlines in the memory array.

[0020] During the program operation the voltages applied to the wordline 216 and the bitline 212 are reversed, with V_POS applied to the wordline and V_NEG applied to the bitline, to apply a bias to program the memory transistor 202. The substrate connection 206 or connection to the well in which the memory transistor 202 is formed is coupled to electrical ground, V_NEG or to a voltage between ground and V_NEG. The read or select line 228 is likewise coupled to electrical ground (0V), and the source line 224 may be at equipotential with the bitline 212, i.e., coupled to V_NEG, or allowed to float.

[0021] After an erase operation or program operation is completed, the state of the memory cell 200 can be read by setting a gate-to-source voltage (V_GS) of the memory transistor 202 to zero, applying a small voltage between the drain terminal 210 and source terminal 218, and sensing a current that flows through the memory transistor. In the programmed state, an N-type SONOS memory transistor, for example, will be OFF because V_GS will be below the programmed threshold voltage V_TP. In the erased state, the N-type memory transistor will be ON because the V_GS will be above an erased threshold voltage V_TE. Conventionally, the ON state is associated with a logical "0" and the OFF state is associated with a logical "1."

[0022] A memory array of memory cells and methods of operating the same to reduce disturbs will now be described with reference to FIG. 3 and Table I below. In the following description, for clarity and ease of explanation, it is assumed that all of the transistors in memory array are N-type SONOS transistors. It should be appreciated, without loss of generality that a P-type configuration can be described by reversing the polarity of the applied voltages, and that such a configuration is within the contemplated embodiments of the invention. In addition, the voltages used in the following description are selected for ease of explanation and represent only one exemplary embodiment of the invention. Other voltages may be employed in different embodiments of the invention.

[0023] FIG. 3 illustrates an exemplary embodiment of a segment of a memory array 300, which may be part of a large memory array of memory cells. In FIG. 3, memory array 300 includes four memory cells 301, 302, 303 and 304 arranged in two rows (ROW 1, ROW 2) and two columns (COLUMN 1, COLUMN 2). Each of the memory cells 301-304 may be structurally equivalent to memory cell 200 described above.

[0024] Referring to FIG. 3, memory cell 301 is the targeted cell to be programmed to a logic "1" state (i.e., programmed to an ON state) while memory cell 302, already erased to a logic "0" state by a preceding erase operation, is maintained in a logic "0" or OFF state. These two objectives (programming cell 301 and inhibiting cell 302) are accomplished by applying a first or positive high voltage (V_POS) to a first global wordline (GWL₁) in the first row of the memory array 300, a second or negative high voltage (V_NEG), is applied to a first bitline (BL₁) to bias transistor T1 on programming the selected memory cell 301, while an inhibit voltage (V_Inhib) is applied to a second bitline (BL₂) to bias transistor T2 off on inhibiting programming of the deselected memory cell 302, and a common or shared voltage is applied to the substrate nodes (SUB) of all memory cells 301, 302, 303 and 304, and the read lines (RL1 and RL2) coupled to electrical ground (0V). The source lines (SL1 and SL2) may be at equipotential with the bitlines in their respective columns, i.e., SL1 is coupled to V_NEG and SL2 coupled to the V_Inhib, or allowed to float.

[0025] In addition, and as described in greater detail below, a selected margin voltage (V_MARG) having a voltage level or magnitude less than V_NEG is applied to a second global wordline (GWL₂) in the second row of the memory array 300 to reduce or substantially eliminate program-state bitline disturb in the deselected memory cell 304 due to programming of the selected memory cell 301.

[0026] Table I depicts exemplary bias voltages that may be used for programming a non-volatile memory having a 2T-architecture and including memory cells with N-type SONOS transistors.

Table I

GWL1	BL1	SL1	RL1	Substrate Node	GWL2	BL2	SL2	RL2
VPOS +4.7V	VNEG -3.6V	Float/ -3.6V	VGND 0.0V	VNEG -3.6V	VMarg -2.6V	VInhib +1.2V	Float/ +1.2V	VGND 0.0V

[0027] Because the voltage applied to the second global wordline (GWL2) has a lower voltage level or magnitude that V_NEG, which is conventionally applied to wordlines in deselected row or cells, the gate to drain voltage (V_GD) across transistor T4 is 3.8V, as compared to a V_GD in conventionally operated memories of 4.8V, the amount of bitline disturb of the threshold V_T of T4 is reduced significantly. In one embodiment of this invention it was observed to be reduced from about 60mV to less than about 7mV.

[0028] The margin voltage (V_MARG) can be generated using dedicated circuitry in the memory (not shown in this figure) used solely for generating V_MARG, or can be generated using circuitry already included in the memory device. Generally, the margin voltage (V_MARG) has the same polarity as the second or V_NEG high voltage, but is higher or more positive than V_NEG by a voltage equal to at least the threshold voltage (V_T) of the transistor T4 in the memory cell 304 for which program state bitline disturb is reduced. Optionally, the circuitry used to generate the margin voltage (V_MARG) is programmable to set a desired margin voltage (V_MARG) with steps, in one embodiment, of 14 mV or less.

[0029] In one embodiment, the circuitry used to generate the margin voltage (V_MARG) includes a digital-to-analog-converter (DAC) enabled by command and control circuitry in the memory programmed to generate a margin voltage (V_MARG) of a desired magnitude or voltage level to be coupled to the GWLs of deselected row(s) during the program operation. In one particular advantageous embodiment the DAC is a margin mode DAC in the memory, which is used during initialization of the memory to adjust voltages therein, and which is not normally enabled during the program operation. Significant advantages of this embodiment include that V_MARG can be trimmed using the (MDAC) bits, it does not represent a large load on a negative pump for V_NEG and an output buffer of the margin mode DAC offers a low impedance driver for the V_MARG signal. Adapting such a margin mode DAC for generating V_MARG during the program operation requires forming an electrical connection to the GWLs of deselected rows of the memory array 300 during the program operation, and enabling the margin mode DAC through a DAC enable signal.

[0030] In certain embodiments, further adaption of the V_MARG circuit is desirable to overcome the fact that V_MARG was not originally designed to drive large capacitive loads active during program. One method of overcoming this limitation will now be described with reference to the graphs of FIGs. 4 and 5.

[0031] FIG. 4 is a graph illustrating a positive first high voltage (V_POS 402), a negative second high voltage (V_NEG 404), and an intermediate, margin voltage (V_MARG 406) according to an embodiment of the present disclosure. Referring to FIG. 4 it is noted that the start-up time for the circuit generating the margin voltage (V_MARG 406) can be relatively slow, up to 80-110 µs, as compared to the second high voltage (V_NEG 404). During this time the voltage difference between a deselected global wordline (GWL₂) to which the margin voltage (V_MARG 406) is applied and the p-well (SPW) or substrate node to which second high voltage (V_NEG 404), can reach 1.6 -1.7 volts for 20-40 µs. Thus, to reduce erase-state bitline disturb in an unselected memory cell in the first column and second row of the memory array (e.g., cell T3), V_NEG is coupled to the second global wordline (GWL₂) in the deselected row for up to about 40 µs until a capacitance associated with the deselected wordline(s) is sufficiently pre-charged, and V_NEG has reached a value close to -2.0 volts. The margin voltage is then coupled to the global wordline (GWL₂) in the deselected row for the remainder of the program operation to reduce program-state bitline disturb in a second unselected memory cell in the second column and second row of the memory array due to programming of the selected memory cell.

[0032] A graph illustrating voltages applied to a selected global wordline (V_{SELECTED WL} 502) and a deselected global wordline (V_{DESELECTED GWL} 504) during a program operation according to an embodiment of the present disclosure is shown in FIG. 5. Referring to FIG. 5 it is noted from the graph of the deselected global wordline voltage (V_{DESELECTED GWL} 504) that at about 15 µs, indicated by reference numeral 506 on the graph of the deselected global wordline voltage, the global wordline (GWL₂) in the deselected row is switched from being coupled to second high voltage (V_NEG 404), to being coupled to the margin voltage (V_MARG 406) for the remainder of the program operation.

[0033] A processing system 600 to reduce bitline program disturbs according to an embodiment of the present disclosure will now be described with reference to FIG. 6.

[0034] Referring to FIG. 6 the processing system 600 generally includes a non-volatile memory 602 coupled to a processor 604 in a conventional manner via an address bus 606, a data bus 608 and a control bus 610. It will be appreciated by those skilled in the art that the processing system of FIG. 6 has been simplified for the purpose of illustrating the present invention and is not intended to be a complete description. In particular, details of the processor, row and column decoders, sense amplifiers and command and control circuitry, which are known in the art have are not described in detail herein.

[0035] The processor 604 may be a type of general purpose or special purpose processing device. For example, in one embodiment the processor can be a processor in a programmable system or controller that further includes a non-volatile memory, such as a Programmable System On a Chip or PSoC^™ controller, commercially available from Cypress Semiconductor of San Jose, California.

[0036] The non-volatile memory 602 includes a memory array 612 organized as rows and columns of non-volatile memory cells (not shown in this figure) as described above. The memory array 612 is coupled to a row decoder 614 via multiple wordlines and read lines 616 (at least one wordline and one read line for each row of the memory array) as described above. The memory array 612 is further coupled to a column decoder 618 via a multiple bitlines and source lines 620 (one each for each column of the memory array) as described above. The memory array 612 is coupled to a plurality of sense amplifiers 622 to read multi-bit words therefrom. The non-volatile memory 602 further includes command and control circuitry 624 to control the row decoder 614, the column decoder 618 and sense amplifiers 622, and to receive read data from sense amplifiers. The command and control circuitry 624 includes voltage control circuitry 626 to generate the voltages needed for operation of the non-volatile memory 602, including V_POS, V_NEG and V_INHIB, and a margin mode DAC 628 to generate V_MARG described above, which is routed through the voltage control circuitry to the row decoder 614. The voltage control circuitry 626 operates to apply appropriate voltages to the memory cells during read, erase and program operations.

[0037] The command and control circuitry 624 is configured to control the row decoder 614 to select a first row of the memory array 612 for a program operation by applying a V_POS to a first global wordline (GWL₁) in the first row and to deselect a second row of the memory array by applying a margin voltage to a second global wordline (GWL₂) in the second row. In some embodiments, the command and control circuitry 624 is configured to sequentially couple first V_NEG to the second global wordline for a brief period of time and then the margin voltage. As described above, in some embodiments, the start-up time for a margin voltage circuit can be relatively slow as compared to that of V_NEG coupled to a substrate node or p-well (SPW) in which the memory transistor is formed, and during this time the voltage bias difference between the deselected wordline (GWL₂) and a p-well (SPW) or substrate node can cause erase-state bitline disturb in an unselected memory cell in the first column and second row of the memory array (e.g., cell T3). Thus, to reduce erase-state bitline disturb in the unselected memory cell in the first column and second row of the memory array (e.g., cell T3), V_NEG is coupled to the second global wordline (GWL₂) in the deselected row for a brief time until a capacitance associated with the deselected wordline(s) is sufficiently pre-charged, and V_NEG has reached a value close to -2.0 volts. The margin voltage is then coupled to the global wordline (GWL₂) in the deselected row for the remainder of the program operation to reduce program-state bitline disturb in a second unselected memory cell in the second column and second row of the memory array due to programming of the selected memory cell.

[0038] The command and control circuitry 624 is further configured to control the column decoder 618 to select a memory cell in the first row (e.g., cell T1) for programming by applying a V_NEG to a first shared bitline (BL₁) in a first column, and to inhibit a unselected memory cell in the first row (e.g., cell T2) from programming by applying an inhibit voltage to a second shared bitline (BL₂) in a second column. The column decoder 618 may be further configured to apply V_NEG to a first shared source line (SL₁) in the first column, and to apply the inhibit voltage on a second shared source line (SL₂) in the second column.

[0039] Details of the command and control circuitry of a memory device according to various embodiments of the present disclosure will now be described with reference to FIGs. 7A-7C.

[0040] Referring to FIG. 7A, in one embodiment the command and control circuitry 700 includes a negative HV supply or pump 702 to generate a V_NEG coupled to the bitline and source line of the selected cell, and to the substrate nodes during the program operation, a digital-to-analog-converter (DAC 704) enabled by the command and control circuitry to generate a margin voltage to be coupled to the GWLs of deselected rows during the program operation, and a switching circuit 706 to switch between V_NEG and the margin voltage coupled to the deselected GWLs during the program operation. The DAC 704 can be a dedicated DAC used solely for generating V_MARG, or a DAC already included in the command and control circuitry 700 or voltage control circuitry 626 for other purposes, and which is normally not utilized during a program operation. As noted above, in one particular advantageous embodiment the DAC is a margin mode DAC 628 in the command and control circuitry 624 of the non-volatile memory 602, which is used during test to measure the threshold voltages of the non-volatile devices therein, and which is not normally enabled during the program operation. It will be appreciated that adapting such a margin mode DAC for generating V_MARG during the program operation requires forming an electrical connection to the switching circuit 706, and through the switching circuit and the row decoder (not shown in this figure) to the GWLs of deselected rows of the memory array during the program operation. The command and control circuitry 624 of the non-volatile memory 602 enables the DAC 704 through a DAC enable signal, and, optionally, operates the DAC to provide a programmed margin voltage level or magnitude. Generally, the DAC 704 is operated to provide a margin voltage having a magnitude less than the voltage magnitude of V_NEG, i.e., higher or more positive than V_NEG in the N-type SONOS embodiment described above, by a voltage equal to at least the threshold voltage (V_T) of the of the memory transistor in the memory cell. In other embodiments, the DAC 704 may be programmed or operated to provide a margin voltage magnitude less than V_NEG by an amount close to the V_T of the memory transistor. For example, in one embodiment described above the DAC 704 may be programmed or operated to provide a margin voltage adjustable to within one or more small steps of about 14mV each.

[0041] In another embodiment, shown in FIG. 7B, the command and control circuitry 700 includes a second charge pump 708 to generate the margin voltage to be coupled to the GWLs of deselected rows during the program operation. By selecting the second charge pump 708 to have a start-up time and power to charge the capacitance associated with the deselected wordline(s) that are substantially the same as the negative pump 702, the GWLs of the deselected rows can be coupled to the margin voltage throughout the program operation, and thus the need for a separate switching circuit 706 is eliminated.

[0042] In the present invention, shown in FIG. 7C, the command and control circuitry 700 includes a voltage divider 710 coupled to an output of negative pump 702 to generate the margin voltage to be coupled to the GWLs of deselected rows during the program operation. Because V_NEG and V_MARG are both supplied by the negative pump 702 there is substantially no difference in start-up time between V_NEG and V_MARG and the voltage bias difference between V_MARG applied the deselected wordline (GWL₂) and V_NEG applied to the p-well (SPW) or substrate node cannot reach a voltage level sufficient to cause erase-state bitline disturb in the unselected memory cell in the first column and second row of the memory array (e.g., 1.6 -1.7 volts for 20-40 µs), the GWLs of the deselected rows can be coupled to the margin voltage throughout the program operation, and thus the need for a separate switching circuit 706 is eliminated.

[0043] FIG. 8 is a flowchart illustrating a method for reducing program disturb in one embodiment. Note, it will be understood that although all steps of the method are described individually below implying a sequential order that is not necessarily the case, and that as shown in FIG. 8, a first five individual steps of the method are performed at substantially the same time, while a last two steps are performed in order after only a slight delay.

[0044] Referring to FIG. 8, a first positive high voltage (V_pos) is coupled to a first global wordline in a first row of a memory array of memory cells (802). In the next operation, a V_NEG is coupled to a first shared bitline in a first column of the memory array to apply a bias to a non-volatile memory transistor in a selected memory cell to program the selected memory cell (804). In embodiments in which the memory transistors are formed in wells in a substrate, the wells may be coupled to electrical ground, a voltage between ground and V_NEG, or, as in the embodiment shown to V_NEG (806). Optionally, V_NEG may be coupled to a second global wordline in a second row of the memory array for a brief period of time to apply a bias to a non-volatile memory transistor in a first unselected memory cell in the first column and the second row of the memory array sharing the first shared bitline with the selected memory cell to reduce erase-state bitline disturb in the first unselected memory cell (808). Simultaneously, a margin voltage less than V_NEG is generated (810). In the next operation, after only a slight delay the margin voltage is coupled to the second global wordline in the second row of the memory array

[0045] (812). In the next operation, an inhibit voltage is coupled to a second shared bitline in a second column of the memory array to apply a bias to a non-volatile memory transistor in a second unselected memory cell in the second row and second column to reduce program-state bitline disturb in the second unselected memory cell (814).

[0046] Thus, embodiments of a non-volatile memory and methods of operating the same to reduce disturbs have been described.

Claims

1. A method comprising:

coupling a first positive high voltage (V_POS) to a first global wordline in a first row of a memory array of memory cells, and coupling a second negative high voltage (V_NEG) to a first bitline and a first sourceline in a first column of the memory array to apply a bias to a nonvolatile memory transistor in a selected memory cell to program the selected memory cell; and

coupling a margin voltage (V_MARG) having a magnitude less than V_NEG to a second global wordline in a second row of the memory array, and coupling an inhibit voltage (V_INHIB) to a second bitline in a second column of the memory array to reduce a bias applied to a non-volatile memory transistor in an unselected memory cell to reduce program disturb of data programmed in the unselected memory cell due to programming of the selected memory cell,

wherein the memory cells each comprise at least a non-volatile memory, NVM, transistor, arranged in rows and columns, wherein gates of the NVM transistors of memory cells in a same row couple to and share a global wordline; and

further including at least one select transistor associated with each memory transistor, wherein the select transistor and the memory cell have different gate terminal connections, the gate of the select transistor coupling to a select line;

wherein V_NEG is generated using a charge pump, and wherein coupling the margin voltage to the second global wordline comprises generating the margin voltage using a voltage divider coupled to an output of the charge pump.

2. The method of claim 1 wherein the margin voltage has a magnitude less than V_NEG by at least a threshold voltage (V_T) of a transistor in the unselected memory cell.

3. The method of claim 2, wherein the transistor is the non-volatile memory transistor in the unselected memory cell.

4. The method of claim 2, wherein coupling the margin voltage to the second global wordline comprises generating the margin voltage using a digital-to-analog-converter, DAC.

5. The method of claim 4, wherein the DAC is programmable, and wherein generating the margin voltage comprises programming the DAC to generate a voltage magnitude less than V_NEG.

6. The method of claim 1, wherein the margin voltage is coupled to the second global wordline through a switching circuit configured to switch the second global wordline between V_NEG and the margin voltage.

7. The method of claim 6, wherein coupling the margin voltage to the second global wordline comprises sequentially coupling V_NEG to the second global wordline for a time before coupling the margin voltage to the second global wordline to reduce the bias applied to the non-volatile memory transistor in the unselected memory cell to reduce program disturb of data programmed in the unselected memory cell due to programming of the selected memory cell.

8. The method of claim 7, wherein the non-volatile memory transistors are formed in wells in a substrate, and further comprising coupling V_NEG to the wells, and wherein the time for which V_NEG is coupled to the second global wordline is less than a time required for a voltage of the wells to increase to V_NEG.

9. The method of claim 1, wherein the non-volatile memory transistor comprises a Silicon-Oxide-Nitride-Oxide-Silicon, SONOS, transistor.

Ansprüche

1. Verfahren, umfassend:

Koppeln einer ersten, positiven Hochspannung (Vpos) mit einer ersten globalen Wortleitung in einer ersten Zeile einer Speichermatrix von Speicherzellen und Koppeln einer zweiten, negativen Hochspannung (VNEG) mit einer erster Bitleitung und einer ersten Sourceleitung in einer ersten Spalte der Speichermatrix, um eine Vorspannung an einen nichtflüchtigen Speichertransistor in einer ausgewählten Speicherzelle anzulegen, um die ausgewählte Speicherzelle zu programmieren; und

Koppeln einer Toleranzspannung (VMARG), die eine kleinere Größe als die VNEG aufweist, mit einer zweiten globalen Wortleitung in einer zweiten Zeile der Speichermatrix und Koppeln einer Sperrspannung (VINHIB) mit einer zweiten Bitleitung in einer zweiten Spalte der Speichermatrix, um eine an einen nichtflüchtigen Speichertransistor in einer nicht ausgewählten Speicherzelle angelegte Vorspannung zum Reduzieren von Programmstörung von in der nicht ausgewählten Speicherzelle programmierten Daten aufgrund des Programmierens der ausgewählten Speicherzelle zu reduzieren,

wobei die Speicherzellen jeweils mindestens einen nichtflüchtigen Speichertransistor, NVM-Transistor, umfassen, angeordnet in Zeilen und Spalten, wobei Gates der NVM-Transistoren von Speicherzellen in einer gleichen Zeile mit einer globalen Wortleitung gekoppelt sind und sich diese teilen; und

ferner umfassend mindestens einen Auswahltransistor, der jedem Speichertransistor zugeordnet ist, wobei der Auswahltransistor und die Speicherzelle verschiedene Gateanschlussverbindungen aufweisen, wobei das Gate des Auswahltransistors mit einer Auswahlleitung gekoppelt ist,

wobei die VNEG unter Verwendung einer Ladepumpe erzeugt wird und wobei das Koppeln der Toleranzspannung mit der zweiten globalen Wortleitung ein Erzeugen der Toleranzspannung unter Verwendung eines Spannungsteilers umfasst, der mit einem Ausgang der Ladepumpe gekoppelt ist.

2. Verfahren nach Anspruch 1, wobei die Toleranzspannung eine um mindestens eine Schwellenspannung (VT) eines Transistors in der nicht ausgewählten Zelle kleinere Größe als die VNEG aufweist.

3. Verfahren nach Anspruch 2, wobei der Transistor der nichtflüchtige Speichertransistor in der nicht ausgewählten Speicherzelle ist.

4. Verfahren nach Anspruch 2, wobei das Koppeln der Toleranzspannung mit der zweiten globalen Wortleitung ein Erzeugen der Toleranzspannung unter Verwendung eines Digital-Analog-Wandlers, DAC, umfasst.

5. Verfahren nach Anspruch 4, wobei der DAC programmierbar ist und wobei das Erzeugen der Toleranzspannung ein Programmieren des DAC umfasst, um eine Spannungsgröße zu erzeugen, die kleiner als die VNEG ist.

6. Verfahren nach Anspruch 1, wobei die Toleranzspannung über einen Schaltkreis, der zum Schalten der zweiten globalen Wortleitung zwischen der VNEG und der Toleranzspannung konfiguriert ist, mit der zweiten globalen Wortleitung gekoppelt ist.

7. Verfahren nach Anspruch 6, wobei das Koppeln der Toleranzspannung mit der zweiten globalen Wortleitung ein sequenzielles Koppeln der VNEG mit der zweiten globalen Wortleitung für eine Zeit vor dem Koppeln der Toleranzspannung mit der zweiten globalen Wortleitung umfasst, um die an den nichtflüchtigen Speichertransistor in der nicht ausgewählten Speicherzelle angelegte Vorspannung zum Reduzieren von Programmstörung von in der nicht ausgewählten Speicherzelle programmierten Daten aufgrund des Programmierens der ausgewählten Speicherzelle zu reduzieren.

8. Verfahren nach Anspruch 7, wobei die nichtflüchtigen Speichertransistoren in Mulden in einem Substrat gebildet werden, und ferner umfassend ein Koppeln der VNEG mit den Mulden, und wobei die Zeit, für die die VNEG mit der zweiten globalen Wortleitung gekoppelt wird, kürzer als eine Zeit ist, die eine Spannung der Mulden zum Ansteigen auf die VNEG benötigt.

9. Verfahren nach Anspruch 1, wobei der nichtflüchtige Speichertransistor einen Silizium-Oxid-Nitrid-Oxid-Silizium-Transistor, SONOS-Transistor, umfasst.

Revendications

1. Procédé comprenant les étapes suivantes :

coupler d'une première haute tension positive (V_POS) à une première ligne de mots globale dans une première rangée d'un réseau de cellules de mémoire, et coupler une deuxième haute tension négative (V_NEG) à une première ligne de bits et à une première ligne de sources dans une première colonne du réseau de mémoire pour appliquer une polarisation à un transistor de mémoire non volatile dans une cellule de mémoire sélectionnée afin de programmer la cellule de mémoire sélectionnée ; et

coupler une tension de marge (V_MARG) dont la magnitude est inférieure à V_NEG à une deuxième ligne de mots globale dans une deuxième rangée de la matrice de mémoire, et coupler une tension d'inhibition (V_INHIB) à une deuxième ligne de bits dans une deuxième colonne de la matrice de mémoire pour réduire une polarisation appliquée à un transistor de mémoire non volatile dans une cellule de mémoire non sélectionnée afin de réduire la perturbation des données programmées dans la cellule de mémoire non sélectionnée en raison de la programmation de la cellule de mémoire sélectionnée,

dans lequel les cellules de mémoire comprennent chacune au moins un transistor de mémoire non volatile, NVM, disposé en rangées et en colonnes, dans lequel les portes des transistors NVM des cellules de mémoire d'une même rangée se couplent à une ligne de mots globale et la partagent ; et

et comprennent en outre au moins un transistor de sélection associé à chaque transistor de mémoire, le transistor de sélection et la cellule de mémoire ayant des connexions de borne de grille différentes, la grille du transistor de sélection se couplant à une ligne de sélection ;

dans lequel V_NEG est généré à l'aide d'une pompe de charge, et dans lequel le couplage de la tension de marge à la deuxième ligne de mots globale comprend de générer la tension de marge à l'aide d'un diviseur de tension couplé à une sortie de la pompe de charge.

2. Procédé selon la revendication 1, dans lequel la tension de marge a une magnitude inférieure à V_NEG d'au moins une tension de seuil (V_T) d'un transistor dans la cellule de mémoire non sélectionnée.

3. Procédé selon la revendication 2, dans lequel le transistor est le transistor de mémoire non volatile dans la cellule de mémoire non sélectionnée.

4. Procédé selon la revendication 2, dans lequel le couplage de la tension de marge à la deuxième ligne de mots globale comprend de générer la tension de marge à l'aide d'un convertisseur numérique-analogique, DAC.

5. Procédé selon la revendication 4, dans lequel le DAC est programmable, et dans lequel la génération de la tension de marge comprend de programmer le DAC pour générer une magnitude de tension inférieure à V_NEG.

6. Procédé selon la revendication 1, dans lequel la tension de marge est couplée à la deuxième ligne de mots globale à travers un circuit de commutation configuré pour commuter la deuxième ligne de mots globale entre V_NEG et la tension de marge.

7. Procédé selon la revendication 6, dans lequel le couplage de la tension de marge à la deuxième ligne de mots globale comprend de coupler de manière séquentielle V_NEG à la deuxième ligne de mots globale pendant un temps avant le couplage de la tension de marge à la deuxième ligne de mots globale pour réduire la polarisation appliquée au transistor de mémoire non volatile dans la cellule de mémoire non sélectionnée pour réduire la perturbation des données programmées dans la cellule de mémoire non sélectionnée en raison de la programmation de la cellule de mémoire sélectionnée.

8. Procédé selon la revendication 7, dans lequel les transistors de mémoire non volatile sont formés dans des puits d'un substrat, le procédé comprenant en outre de coupler V_NEG aux puits, et dans lequel le temps pendant lequel V_NEG est couplé à la deuxième ligne de mots globale est inférieur à un temps nécessaire pour qu'une tension des puits augmente jusqu'à V_NEG.

9. Procédé selon la revendication 1, dans lequel le transistor de mémoire non volatile comprend un transistor Silicium-Oxide-Nitrure-Oxide-Silicium, SONOS.

Drawing